Molecular Phylogeny and Biogeographic History of the Armored Neotropical Subfamilies , and Otothyrinae (Siluriformes: )

Fa´bio F. Roxo1*, James S. Albert2, Gabriel S. C. Silva1, Cla´ udio H. Zawadzki3, Fausto Foresti1, Claudio Oliveira1 1 Laborato´rio de Biologia e Gene´tica de Peixes, Departamento de Morfologia, Universidade Estadual Paulista, UNESP, Botucatu, SP, Brazil, 2 Department of Biology, University of Louisiana at Lafayette, Lafayette, Louisiana, United States of America, 3 Nupe´lia, Universidade Estadual de Maringa´, UEM, Maringa´, PR, Brazil

Abstract The main objectives of this study are estimate a -dense, time-calibrated molecular phylogeny of Hypoptopomatinae, Neoplecostominae, and Otothyrinae, which together comprise a group of armoured that is widely distributed across , to place the origin of major clades in time and space, and to demonstrate the role of river capture on patterns of diversification in these taxa. We used maximum likelihood and Bayesian methods to estimate a time-calibrated phylogeny of 115 loricariid species, using three mitochondrial and one nuclear genes to generate a matrix of 4,500 base pairs, and used parametric biogeographic analyses to estimate ancestral geographic ranges and to infer the effects of river capture events on the geographic distributions of these taxa. Our analysis recovered Hypoptopomatinae, Neoplecosto- minae, and Otothyrinae as monophyletic with strong statistical support, and Neoplecostominae as more closely related to Otothyrinae than to Hypoptopomatinae. Our time-calibrated phylogeny and ancestral-area estimations indicate an origin of Hypoptopomatinae, Neoplecostominae, and Otothyrinae during the Lower Eocene in the Atlantic Coastal Drainages, from which it is possible to infer several dispersal events to adjacent river basins during the Neogene. In conclusion we infer a strong influence of river capture in: (1) the accumulation of modern clade species-richness values; (2) the formation of the modern basin-wide species assemblages, and (3) the presence of many low-diversity, early-branching lineages restricted to the Atlantic Coastal Drainages. We further infer the importance of headwater stream capture and marine transgressions in shaping patterns in the distributions of Hypoptopomatinae, Neoplecostominae and Otothyrinae throughout South America.

Citation: Roxo FF, Albert JS, Silva GSC, Zawadzki CH, Foresti F, et al. (2014) Molecular Phylogeny and Biogeographic History of the Armored Neotropical Catfish Subfamilies Hypoptopomatinae, Neoplecostominae and Otothyrinae (Siluriformes: Loricariidae). PLoS ONE 9(8): e105564. doi:10.1371/journal.pone.0105564 Editor: Helge Thorsten Lumbsch, Field Museum of Natural History, United States of America Received May 14, 2014; Accepted July 21, 2014; Published August 22, 2014 Copyright: ß 2014 Roxo et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability: The authors confirm that all data underlying the findings are fully available without restriction. Data are available from the GenBank sequence database and accession numbers are listed in the supporting information. Funding: This research was supported by the Brazilian agency FAPESP (Fundac¸a˜o de Amparo a` Pesquisa do Estado de Sa˜o Paulo, proc. 2010/01610-9 to FFR and proc. 2012/01622-2 to GSCS). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing Interests: The authors have declared that no competing interests exist. * Email: [email protected]

Introduction river drainage basin is diverted into that of a neighbouring basin. River capture may arise from the influence of several geomor- A central aim of research in modern historical biogeography is phological processes, including tectonic uplift or tilting, damming to understand the distributions of species and ecosystems in light of by the actions of glaciers or landslides, denudation of watershed historical processes that shape landscape evolution [1,2]. This margins by erosion, or avulsion of watershed margins by sediment effort has made rapid progress over the past decade in the study of accumulation in fans and estuaries [12–16]. The process of Neotropical freshwater . The continental fishes of tropical headwater or lateral watershed erosion has the remarkable effect South America represent about one fifth of the world’s species, of simultaneously separating portions of river basins that were or 10% of all vertebrate species [3,4]. The evolutionary and formerly connected (i.e., vicariance), and of connecting portions of ecological reasons for the origins and maintenance of this high river basins that were formerly isolated (i.e., geodispersal). In other diversity remain poorly understood. However, for obligate words, river capture moves the physical location of watershed freshwater organisms, river capture is an important landscape- barriers [7,17]. For obligate aquatic taxa, such as freshwater fishes, level process that can isolate lineages and promote diversification amphibians, and other and plant groups that inhabit [5–7] by changing patterns in the connectivity of adjacent river riparian or floodplain habitats, river capture facilitates the basins [8–11]. dispersal of species between adjacent drainage basins. River capture (also called stream capture or stream piracy) is a The biogeographic consequences of river capture can be geomorphological process by which the flow of part of a stream or profound for species that are restricted to river basins, and for

PLOS ONE | www.plosone.org 1 August 2014 | Volume 9 | Issue 8 | e105564 Molecular Phylogeny and Biogeographic History of Cascudinhos which watershed boundaries strongly limit dispersal [17–19]. As in Astroblepus sp. 2 (Astroblepidae), gobio, H. all vicariance events, the separation of formerly adjacent river papillatus, Delturus parahybae (Loricariidae, subfamily Deltur- basin segments promotes allopatric divergence, and may ultimate- inae), Rineloricaria lanceolata, Spatuloricaria sp. 1 (Loricariidae, ly lead to speciation. However, in addition, and unlike many subfamily ), Hypostomus ancistroides, H. nigromacu- vicariance-only events, river capture always results in both the latus and H. microstomus (Loricariidae, subfamily Hypostominae) separation and the merging of adjacent river basin segments [4,20]. were included in the analysis as additional outgroups. We included In other words, in the special case of river capture, vicariance and in the analysis 155 specimens representing 115 loricariid species geodispersal are near simultaneous and complementary biogeo- (see table S1 to all species names, localities and deposits in graphic processes. Further, both vicariance and geodispersal may museums and table S2 to taxonomic summary of ingroup species). result in concordant biogeographic patterns among many lineages that constitute a regional biota, (see fig. 10 in Lieberman [21], Ethics Statement [22]). Importantly, in the case of river capture, and under the All fishes collected for this study were collected in accordance widely-used convention that geographic range is defined in terms with Brazilian laws, under a permanent scientific collection license of the river basin(s) a species occupies (e.g. [23,24]), geodispersal in the name of Dr. Claudio Oliveira (SISBIO). Additionally, our can result in geographic range expansion without necessarily laboratory has special federal permission to keep and involving biotic dispersal [25]. tissues from a public collection under our care. To work with the Among Neotropical freshwater fishes, members of the family animals, we follow all the ethical prescriptions stated by our Loricariidae, armoured catfishes, represent an excellent model to internal committee of ethic called ‘‘Comissa˜o de E´ tica na study the effects of landscape evolution on lineage diversification. Experimentac¸a˜o Animal’’ (CEEA) involving animal experiments Loricariids inhabit most aquatic habitats and geographic regions (protocol number 388) that approved this study. After collection, of tropical South and Central America. About 869 loricariid animals were anesthetized with benzocaine, and a piece of muscle species are currently recognized as valid [26], which makes this tissue was extracted from the right side of the body and preserved taxon the second-most species-rich family of Neotropical fresh- in 95% ethanol. Voucher specimens were fixed in 10% formalin water fishes (after Characidae). Loricariids also exhibit a broad for two weeks, and then transferred to 70% ethanol for permanent range of ecological tolerances and geographic distributions. Many storage. species are extreme habitat or trophic specialists [27–31], and Vouchers of all samples were deposited in the collection of the many species are highly endemic, with small geographic ranges Laborato´rio de Biologia e Gene´tica de Peixes (LBP), Departa- [16,22,33]. mento de Morfologia, Instituto de Biocieˆncias, Universidade Within the Loricariidae the three subfamilies, - Estadual Paulista, Botucatu, Sa˜o Paulo, Brazil, Museu de Cieˆncias tinae, Neoplecostominae and Otothyrinae, have long been e Tecnologia, Pontifı´cia Universidade Cato´lica do Rio Grande do recognized together as natural group using morphological and Sul (MCP), Porto Alegre, Rio Grande do Sul, Brazil; Nu´cleo molecular data [32,34–38]. Separately, these three subfamilies de Pesquisas em Limnologia, Ictiologia e Aquicultura (NUP), were hypothesized to form monophyletic groups in Chiachio et al. Universidade Estadual de Maringa´, Parana´, Brazil, or the [36] using molecular data. No formal infrafamily name has yet Museum of Natural History of the City of Geneva (MHNG), been applied to this clade, which we will refer to here as the HNO- Geneva, Switzerland. clade. Each of these three subfamily-level clades is also species- rich, and the HNO-clade as a whole has 182 nominal species currently recognized [39]. Each of these subfamilies exhibits a DNA Extraction and Sequencing wide geographic distribution throughout tropical cis-Andean Total DNA was extracted from ethanol preserved muscle South America, and has a lengthy and complex taxonomic samples with the DNeasy Tissue Kit (Qiagen), following manu- history, including studies using both morphological and molecular facturer’s instructions. Partial sequences of the genes 16S rRNA datasets [32,36–38,40–49]. [57], cytochrome b (Cytb) [58], cytochrome c oxidase subunit I Here, we present a time-calibrated phylogenetic analysis of the (COI) [59] and F-reticulon 4 [36] were amplified using polymerase loricariid catfish subfamilies Hypoptopomatinae, Neoplecostomi- chain reaction (PCR) with the primers described in Table S3. nae and Otothyrinae, using a combination of three mitochondrial Amplifications were performed in a total volume of 12.5 ml with and one nuclear gene markers, and the most species-dense taxon 1.25 mlof106 buffer (10 mM Tris-HCl+15 mM MgCl2), 0.5 ml sampling of these groups to date. We then use parametric dNTPs (200 nM of each), 0.5 ml each 5 mM primer, 0.05 ml biogeographic methods to estimate ancestral geographic ranges, Platinum Taq Polymerase (Invitrogen), 1 ml template DNA and to document several historical river-capture events in the (12 ng), and 8.7 ml ddH2O. The PCR reactions consisted of region of Southeastern Brazil. Our results highlight the special role 30–40 cycles, 30 s at 95uC, 15–30 s at 48–58uC (according to of river capture in the formation of the modern species richness primer and species), and 45–90 s at 72uC (according to gene and geographic distributions of the Hypoptopomatinae, Neople- primers). Nested-PCRs were used to amplify the nuclear marker; costominae and Otothyrinae. the first amplification was performed using the primers Freticul4- D and Freticul4-R with a total volume of 12.5 ml for 30–40 cycles Materials and Methods (30 s at 95uC, 30 s at 48uC, and 135 s at 72uC); the second amplification was performed using the primers Freticul4 D2 and Taxon Sampling Freticul4 R2 with a total volume of 12.5 ml for 30–40 cycles (30 s Diplomystes mesembrinus (Diplomystidae) was used as a distant at 95uC, 30 s at 53–54uC, and 135 s at 72uC). All PCR products outgroup to root all phylogenies. Diplomystidae has been were first visually identified on a 1% agarose gel and then purified alternatively been reported as the sister group to all other catfishes, using ExoSap-IT (USB Corporation) following instructions of the or as the sister group to Siluroidea, a clade of catfishes that manufacturer. The purified PCR products were sequenced using excludes Loricarioidea [50–56]. Additionally, samples of Cory- the ‘‘Big DyeTM Terminator v 3.1 Cycle Sequencing Ready doras imitator, Corydoras oiapoquensis, Hoplosternum littorale, Reaction Kit’’ (Applied Biosystems), purified again by ethanol Callichthys callichthys (Callichthyidae), Astroblepus sp. 1 and precipitation and loaded on an automatic sequencer 3130-Genetic

PLOS ONE | www.plosone.org 2 August 2014 | Volume 9 | Issue 8 | e105564 Molecular Phylogeny and Biogeographic History of Cascudinhos

Analyzer (Applied Biosystems) in the Instituto de Biocieˆncias, Expected Likelihood Weights (ELW) method [77]. All tests were Universidade Estadual Paulista, Botucatu, Sa˜o Paulo. conducted under ML with a 14 partition scheme and with the same model implemented in RAxML analysis (Table S4). Sequence and Phylogenetic Analysis All individual sequences for each species were initially analysed Time Calibration and Ancestral-area estimation using the software program BioEdit 5.0.9 [60] and consensus The uncorrelated relaxed molecular clock (lognormal) was sequences were obtained. All sequences for each gene were estimated using BEAST v.1.7.5. All clade-age estimates are independently aligned using MUSCLE [61] under default presented as the mean and 95% highest posterior density (HPD) parameters and the alignments inspected by eye for any obvious values, which are the upper and lower bounds of the HPD misalignments. Only alignment errors were corrected, where interval. We included two calibration points to constrain indels of 1 bp were added to introns of the reticulon gene. A divergence dates for the 154 clades identified in our phylogenetic quality control step was included in our workflow to detect tree. The first calibration point was implemented as a normally- potential cases of sequencing errors due to contamination or distributed prior, with an offset of 125 million years ago (Mya), and paralogy. Alignments for each gene were initially analysed by a standard deviation of 15 million years. These date-estimate maximum likelihood (ML) [62] using the web servers RAxML parameters were selected to match current knowledge of the BlackBox [63] for a previous phylogenetic analysis, and controlling timing of siluriform origins and were implemented in the root of potential sequencing errors involving pseudogenes, paralogous the tree. Information from the stratigraphic record and geographic copies or even laboratory cross-contamination or mistakes during distributions of living taxa indicate an origin for Siluriformes as a the sequencing process. After that, sequences of all genes were whole during the Lower Cretaceous (145–100 Mya; [56,78,79]). concatenated to perform all phylogenetic and biogeography The second calibration point was implemented using a log- analysis (e.g. all four genes for all specimens). normal prior offset to 55 Mya with a mean and standard deviation Sequences that were found misplaced in the resulting gene tree of 1 for the origin of the Corydoras lineage (node including (as, for example, species of one subfamily grouped with species of Corydoras imitator, Corydoras oiapoquensis, Hoplosternum littorale an obviously non-related subfamily) were re-sequenced. Nucleo- and Callichthys callichthys). The oldest known callichthyid fossil, tide variation, substitution patterns, and genetic distances were Corydoras revelatus [80] was dated by Marshall et al. [81] as examined using MEGA 5.0 [64]. To evaluate the occurrence of Paleocene. This prior assumed 55 Mya as a minimum age. We substitution saturation for each gene separately, we estimated the used a macroevolutionary Birth–Death model for the diversifica- index of substitution saturation (Iss) as described by Xia et al. [65] tion likelihood values and a starting tree obtained from the and Xia and Lemey [66] and the rate of transitions/transversions, RAxML analysis. The analyses were conducted under different both evaluated in the software DAMBE 5.2.31 [67]. The Iss models of molecular evolution for each partition of the data matrix estimation was performed without taking into account gaps as evaluated by the software PartitionFinder [68] (Table S4). The because unresolved sites reduce the ability of the method to test ML tree obtained was used as a starting tree for the MCMC for phylogenetic signal. The best-fit partitioning schemes and the searches. The analysis was run for 50 million generations and best nucleotide evolution model for each partition were evaluated sampled every 1000th generation. Stationarity and sufficient in the software PartitionFinder [68] under the information- mixing of parameters (ESS.200) was checked using Tracer v1.5 theoretic measure of Akaike Information Criterion (AICc). [72]. A consensus tree was built using TreeAnnotator v1.7.5 [82]. Maximum likelihood analyses were performed using RAxML Data on the geographic distributions of species in each of the Web-Servers [62]. RAxML implements a faster algorithm of three subfamilies analysed here (Hypoptopomatinae, Neoplecos- heuristic searches with bootstrap pseudoreplicates (RBS). Boot- tominae and Otothyrinae) were taken from the original species strap (BS) resampling [69] was applied to assess support for individual nodes using 1,000 replicates. Random starting trees descriptions and information available at catalog of Eschmeyer were used for each independent ML tree search and all other [26]. We assigned taxa to geographic areas using the ecoregion parameters were set on default values. The ML analyses were classifications of Vari and Malabarba [3] and Chiachio et al. [36], conducted under different models for each partition of the matrix within the following five biogeographic regions: A, Atlantic as evaluated for the software PartitionFinder [68] (Table S4). Coastal Drainages of Southeastern Brazil; B, Upper Parana´ Basin; Bayesian inference (BI) [70] was performed evaluating alternative C, Paraguay, Lower Parana´ and Uruguay basins; D, Amazon and tree topologies through the estimation of posterior probabilities (P) Orinoco basins; E, Sa˜o Francisco basin and Coastal Drainages of using MrBayes v.3.0 [71]. The Bayesian inference was conducted Northeastern of Brazil. under different models for each partition of the matrix as A maximum-likelihood analysis of biogeographic history was evaluated for the software PartitionFinder [68] (Table S4). The also performed in Lagrange v2.0 [(Ree et al., 2005; Ree and ML tree was used as a starting tree for the Markov chain Monte Smith, 2008) using a DEC model of geographic range evolution. Carlo searches. Eight chains were run simultaneously for The DEC model specifies instantaneous transition rates between 100,000,000 generations and every 1000th generation a tree was discrete distribution areas along the branches of a phylogenetic sampled. The above analysis was performed twice. The distribu- tree, and uses these rates to assess the likelihoods of ancestral tion of log-likelihood scores was examined to determine stationary distributions at cladogenetic events [83,84]. Four DEC models phase for each search and to decide if extra runs were required to were tested to estimate distribution ranges inherited by the achieve convergence, using the program Tracer 1.5 [72]. All descending lineages at each node of the tree. The differences sampled topologies beneath the asymptote (25,000,000 genera- between the models are in the rate of dispersal among adjacent tions) were discarded as part of a burn-in procedure, and the and no adjacent areas (see Table S5 for the likelihood values and remaining trees were used to construct a 50% majority-rule dispersal rate among adjacent and no adjacent areas for each consensus tree in Paup* [73]. model). The model that obtained the highest ML values was model Alternative tree topologies were evaluated in the program 3 (M3) that constrained the dispersal rates between adjacent areas Treefinder [74] using the Shimodaira and Hasegawa (SH) test at 0.5 and areas separated by one or more intercalated areas at [75], the Approximately Unbiased (AU) test [76], and the 0.0001.

PLOS ONE | www.plosone.org 3 August 2014 | Volume 9 | Issue 8 | e105564 Molecular Phylogeny and Biogeographic History of Cascudinhos

Figure 1. Partial ML tree showing outgroups and interrelationship among species of the subfamily Hypoptopomatinae. Numbers above branches are bootstrap values from 1000 bootstrap pseudoreplicates obtained from ML analysis. Bootstrap values below 50% (2) are not shown. Numbers below branches are posterior probabilities obtained in the BI analysis. Posterior probabilities values below 0.95 (2) or when the nodes were not obtained by B analyses are not shown. doi:10.1371/journal.pone.0105564.g001

Results (BS = 97, P = 1.00). Tree topology tests rejected the hypothesis that Otothyrinae and Hypoptopomatinae are sister groups Phylogenetic Analysis (as proposed by Schaefer 1991 and 1998) in two (ELW and AU) Partial sequences of three mitochondrial genes (16S rRNA, of the three tests performed (Table S7). The hypothesis that COI, Cytb) and one nuclear gene (F-reticulon 4) were obtained Otothyrinae and Hypoptopomatinae are sister taxa was not from 155 specimens representing 115 loricariid species (Table S1). supported by the SH test, but this test is considered less reliable The combined sequence data resulted in a matrix of exactly 4,500 than the AU test for the same datasets [76]. base pairs (bp), of which 1,482 bp (33%) were non-variable Within Hypoptopomatinae all examined genera were recovered (conserved), 2,677 bp (59%) were variable and included in the as monophyletic with strong statistical support values (BS = 69, analysis, and 341 bp (8%) were variable indels excluded from the P = 1.0 for ; BS = 97, P = 1.0 for ; BS = 100, analysis. This matrix was used to perform all phylogenetic and P = 1 for ; BS = 100, P = 1 for Hypoptopoma). Otocin- biogeographic analyses and was partitioned by gene and coding clus was recovered as the sister group of Lampiella gibbosa, and positions into 14 sections (Table S4). These data were not these taxa together formed the sister group to a clade consisting of saturated considering that the Iss.c value is greater than the Iss, Oxyropsis, Acestridium and Hypoptopoma. Acestridium and 2 and the R value is greater than 0.70 for transitions and Hypoptopoma group together as the sister group to Oxyropsis. transversions for all the genes (Table S6). Within Neoplecostominae and Bayesian and ML phylogenetic analyses resulted in very similar were recovered as monophyletic with high statistical support topologies (Figs. 1–4). Our results illustrate that the clades (BS = 100, P = 1.0 for Kronichthys; BS = 69, P = 0.99 for Isbrueck- Hypoptopomatinae, Neoplecostominae and Otothyrinae are erichthys), however , and Neoplecosto- monophyletic with strong statistical support (BS = 96, P = 0.99 mus were not recovered as monophyletic. The topology tests for Hypoptopomatinae; BS = 99, P = 1.00 for Neoplecostominae; rejected the hypothesis of a monophyletic and BS = 96, P = 0.99 with BI for Otothyrinae). Additionally, our Pareiorhina (Table S7). Pareiorhaphis splendens formed the sister results suggest that Neoplecostominae is more closely related to group to species of Kronichthys, and this group formed the sister Otothyrinae than to Hypoptopomatinae (BS = 98, P = 0.99), and taxon to other species of Pareiorhaphis. that these two clades together form the sister group to Within Otothyrinae , Schizolecis, Rhinolekos, Hypoptopomatinae to the exclusion of other Loricariidae and were monophyletic with high

PLOS ONE | www.plosone.org 4 August 2014 | Volume 9 | Issue 8 | e105564 Molecular Phylogeny and Biogeographic History of Cascudinhos

Figure 2. Partial ML tree showing interrelationship among species of the subfamily Neoplecostominae. Numbers above branches are bootstrap values from 1000 bootstrap pseudoreplicates obtained from ML analysis. Bootstrap values below 50% (2) are not shown. Numbers below branches are posterior probabilities obtained in the BI analysis. Posterior probabilities values below 0.95 (2) or when the nodes were not obtained by B analyses are not shown. doi:10.1371/journal.pone.0105564.g002 statistical support (BS = 100, P = 1.0 for Corumbataia; BS = 100, the species chromodontus, Hisonotus sp. 1, Hisonotus sp. P = 1.0 for Schizolecis; BS = 100, P = 1.0 with BI for Rhinolekos; 2, Hisonotus sp. 3, aripuanensis, Parotocinclus aff. BS = 92, P = 1.0 for Epactionotus, BS = 85, P = 0.99 for Pseudo- spilurus, and Parotocinclus sp. 3. The third lineage is composed of tothyris). The genera Hisonotus and Parotocinclus were not Hisonotus depressicauda, H. francirochai and H. paulinus, and is monophyletic. There are four lineages within the subfamily supported by high statistical support values (BS = 99, P = 1.0). The Otothyrinae that include species currently assigned to Hisonotus. fourth lineage is composed of the most number of Hisonotus The first lineage includes the species Hisonotus insperatus, H. species in this analysis, including Hisonotus aky, H. iota, H. oliveirai, H. paresi, H. piracanjuba, Hisonotus sp. 4, and montanus, H. megaloplax, H. prata, H. carreiro, H. ringueleti, H. Hisonotus sp. 5 and is supported by high statistical support values nigricauda, H. notopagos, H. cf. charrua, H. laevior, H. charrua, (BS = 100 with ML and P = 1). The second lineage is composed of H. leucophrys, H. leucofrenatus, H. cf. taimensis, H. notatus and

PLOS ONE | www.plosone.org 5 August 2014 | Volume 9 | Issue 8 | e105564 Molecular Phylogeny and Biogeographic History of Cascudinhos

Figure 3. Partial ML tree showing interrelationship among species of the subfamily Otothyrinae. Numbers above branches are bootstrap values from 1000 bootstrap pseudoreplicates obtained from ML analysis. Bootstrap values below 50% (2) are not shown. Numbers below branches are posterior probabilities obtained in the BI analysis. Posterior probabilities values below 0.95 (2) or when the nodes were not obtained by B analyses are not shown. doi:10.1371/journal.pone.0105564.g003

H. armatus, and for species marapoama, Eurychei- using BEAST is 0.272% per MY. The Hypoptopomatinae is lichthys sp. 1, Epactionotus bilineatus, E. itaimbezinho and E. estimated by BEAST to have originated during the Lower Eocene gracilis, and is supported by high statistical support values about 33.6–67.4 Mya 95% HPD (mean 49.9 Mya), and is inferred (BS = 72, P = 0.99). by Lagrange to have originated in the Atlantic Coastal Drainages region (Fig. 5, Region A). The clade composed of Neoplecosto- Relaxed Clocks and Historical Biogeography minae + Otothyrinae is estimated by BEAST to also have Our time tree (Figs. 5–7) is the most comprehensive study of originated during the Lower Eocene about 31.0–62.2 Mya 95% catfishes of Hypoptopomatinae, Neoplecostominae and Otothyr- HPD (mean 45.9 Mya), and is also inferred by Lagrange to have inae to date, including 115 loricariid species in these three originated in the Atlantic Coastal Drainages region (Fig. 5, Region A). subfamilies. The mean substitution rate for the dataset estimated

PLOS ONE | www.plosone.org 6 August 2014 | Volume 9 | Issue 8 | e105564 Molecular Phylogeny and Biogeographic History of Cascudinhos

PLOS ONE | www.plosone.org 7 August 2014 | Volume 9 | Issue 8 | e105564 Molecular Phylogeny and Biogeographic History of Cascudinhos

Figure 4. Partial ML tree showing the interrelationship among species of the subfamily Otothyrinae. Numbers above branches are bootstrap values from 1000 bootstrap pseudoreplicates obtained from ML analysis. Bootstrap values below 50% (2) are not shown. Numbers below branches are posterior probabilities obtained in the BI analysis. Posterior probabilities values below 0.95 (2) or when the nodes were not obtained by B analyses are not shown. doi:10.1371/journal.pone.0105564.g004

Hypoptopomatinae is distributed across three of the geographic Basins (D) about 19.8–42.6 Mya 95% HPD (mean 31.2 Mya). regions in Fig. 5: Atlantic Coastal Drainages (Region A), The ancestor of the group composed of the species Microlepido- Paraguay, Lower Parana´ and Uruguay Basins (Region C), and gaster dimorpha, Rhinolekos sp. 1, R. britskii and R. garavelloi Amazon and Orinoco Basins (Region D). The ancestral-area dispersed from the Coastal Drainages region (A) to the Upper estimations suggest that the Hypoptopomatinae moved from Parana´ Basin (B) about 18.3–37.3 Mya 95% HPD (mean Coastal Drainages (Region A) to the Amazon and Orinoco Basins 27.3 Mya). The ancestor of the clade composed of Hisonotus (Region D) between 26.7–58.6 Mya 95% HPD (mean 42.5 Mya). depressicauda, H. francirochai and H. paulinus originated in The relationships among hypoptopomatine taxa in the Amazon Upper Parana´ Basin (B) about 15.7–33.2 95% HPD (mean and Paraguay basins also suggest that these two regions were either 23.7 Mya). The ancestor of the clade composed of Parotocinclus connected or exchanged headwaters at about 15 Mya (Fig. 6). sp. 1, Parotocinclus cf. bahiensis, P. robustus and P. prata, New Neoplecostominae is also distributed across three of the regions taxon sp. 1, New taxon sp. 2 and Hisonotus bocaiuva, originated in in Fig. 5: Coastal Drainages (Region A), Upper Parana´ Basin the Coastal Drainages region (A) about 15.7–33.2 Mya 95% HPD (Region B) and Sa˜o Francisco Basin and Northeastern Drainages (mean 23.7 Mya). The ancestral of the clade composed of the most (Region E). The ancestral lineage of Neoplecostomus (except N. species of Hisonotus and it type species H. notatus its closest ribeirensis), Pareiorhina carrancas is inferred to have reached the relatives (Hisonotus aky, H. iota, H. montanus, H. megaloplax, H. Upper Parana´ Basin from the Coastal Drainages at c. 14.2– prata, H. carreiro, H. ringueleti, H. nigricauda, H. notopagos, H. 33.4 Mya 95% HPD (mean 22.9 Mya). The ancestor of cf. charrua, H. laevior, H. charrua, H. leucophrys, H. leuco- Pareiorhina carrancas reached the Upper Parana´ Basin from frenatus, H. cf. taimensis, H. notatus and H. armatus), and the the Coastal Drainages at c. 1.7–8.7 Mya 95% HPD (mean species Otothyropsis marapoama, sp. 1, Epactio- 4.6 Mya). The ancestor of Neoplecostomus franciscoensis reached notus bilineatus, E. itaimbezinho and E. gracilis originated in the the Sa˜o Francisco basin from Coastal Drainages at c. 3.9– Coastal Drainage region about 17.0–35.0 Mya 95% HPD (mean 13.1 Mya 95% HPD (mean 7.5 Mya). The ancestor of Pseudo- 25.6 Mya). tocinclus tietensis reached the Upper Parana´ Basin (B) from the Additionally, two important dispersal events can be inferred Coastal Drainages region (A) about 0.4–5.5 Mya 95% HPD (mean from the Coastal Drainages region (A) to the Paraguay, Lower 2.3 Mya). The ancestral lineage of Pareiorhaphis eurycephalus, P. Parana´ and Uruguay Basins (C). The first is the ancestor of hystrix, P. parmula and P. vestigipinnis reached the Uruguay Hisonotus iota, H. aky, H. montanus, H. megaloplax, H. prata, H. Basins about 2.0–7.5 Mya 95% HPD (mean 4.3 Mya) (Fig. 6). carreiro and H. ringueleti about 10.7–26.3 Mya 95% HPD (mean The ancestral-area estimations (Fig. 7) suggest that Otothyrinae 17.8 Mya). The second is the ancestor of Hisonotus cf. charrua, originated in the Atlantic Coastal Drainages (Region A) and then H. leucophrys, H. charrua, H. laevior, H. cf. taimensis, subsequently expanded its range into the other regions by means H. leucofrenatus, H. notatus and H. armatus about 11.4–24.8 of biotic dispersal, geodispersal (river capture), or both. The first 95% HPD (mean 17.5 Mya). group to diverge within Otothyrinae is composed for species of the genus Corumbataia and six species of the genus Hisonotus (H. Discussion insperatus, Hisonotus oliveirai, Hisonotus paresi, H. piracanjuba, Hisonotus sp. 4 and Hisonotus sp. 5). The ancestral lineage of this River capture as a biogeographic process group originated in Coastal Drainages region (A) at 29.0– In this study we used a time-calibrated molecular phylogeny and 57.1 Mya 95% HPD (mean 42.5 Mya). The second group to ancestral-area estimations (Figs. 5–7) with robust taxonomic diverge is composed of , the only known sampling, to document the effects of river capture on the species of Schizolecis. Our results suggest that the ancestor of this diversification of taxa in the HNO-clade of loricariid (armoured) species originated in the Coastal Drainages region (A) about 28.7– catfishes in South America. The results are largely consistent with 55.8 Mya 95% HPD (mean 41.0 Mya). The third group to those of previous studies of loricariids from Southern and diverge within Otothyrinae is composed four species of Hisonotus Southeastern Brazil [36,48,85]. For example, Chiachio et al. (Hisonotus sp. 1, Hisonotus sp. 2, Hisonotus sp. 3 and Hisonotus [36] recovered a similar division of the HNO-clade into two chromodontus) and three species of Parotocinclus (Parotocinclus sp. monophyletic groups, the Hypoptopomatinae and Neoplecosto- 3, P. aripuanensis and Parotocinclus aff. spilurus). The ancestor of minae + Otothyrinae, inferred the ancestor of Hypoptopomatinae this group dispersed from the Coastal Drainages region (A) to the to have inhabited the Amazon basin, and inferred the ancestor of Amazon and Orinoco Basins (D) about 25.6–51.0 Mya 95% HPD Neoplecostominae + Otothyrinae to have inhabited an area now (mean 37.5 Mya). Subsequently, the ancestor of the clade drained by the Upper Parana and part of the Atlantic coastal composed of Hisonotus sp. 1, Hisonotus sp. 2 and Parotocinclus drainages. aff. spilurus reached the Sa˜o Francisco Basin and Northeastern The Atlantic coastal region has a complex and ancient Basins (E) about 19.3–43.2 Mya 95% HPD (mean 30.7 Mya). geological history that traces to the final separation of Africa The fourth group to diverge within Otothyrinae is composed of and South America about 100 million years ago [86–89]. Roxo et species of the genus Pseudotothyris, , a Otothyrinae al. [48] identified the Coastal Drainages of Southeastern Brazil as unidentified and the species Parotocinclus sp. 2, P. britskii and P. an important area where many lineages of Loricariidae originated, eppleyi. The ancestor of this group originated in the Coastal including the ancestors of Neoplecostominae. Ribeiro [16] Drainages region (A) about 23.3–46.1 Mya 95% HPD (mean described a set of phylogenetic patterns (termed A, B and C) with 33.7 Mya). Subsequently, the ancestor of Parotocinclus sp. 2, P. sister group relationships observed between lineages inhabiting the britskii and P. eppleyi dispersed from to the Amazon and Orinoco Atlantic coastal drainages and inland drainages such as Amazon

PLOS ONE | www.plosone.org 8 August 2014 | Volume 9 | Issue 8 | e105564 Molecular Phylogeny and Biogeographic History of Cascudinhos

PLOS ONE | www.plosone.org 9 August 2014 | Volume 9 | Issue 8 | e105564 Molecular Phylogeny and Biogeographic History of Cascudinhos

Figure 5. Time-calibrated phylogeny for Hypoptopomatinae, Neoplecostominae and Otothyrinae. Tree topology from BEAST analysis of 155 specimens representing 113 loricariid species. Divergence ages calibrated by origins of Siluriformes (120 Mya) and Callichthyidae (55 Mya). Regions: A, Atlantic Coastal Drainages (Green); B, Upper Parana´ Basin (Red); C, Paraguay, Lower Parana´ and Uruguay Basins (Purple); D, Amazon and Orinoco Basins (Blue); E, Sa˜o Francisco Basin and Northeastern Drainages (Yellow). doi:10.1371/journal.pone.0105564.g005 and Parana´ Basins. Our results suggest a fit to pattern B in important river capture event at this approximate time and place Neoplecostominae and Otothyrinae, with sister-group relation- is also consistent with the topology of a General Area Cladogram ships between species endemic to the Brazilian coastal drainages of fish taxa from tropical South America, as inferred from a Brooks and adjacent portions of the Brazilian shield. Ribeiro [16] listed Parsimony (meta) Analysis of all 32 published phylogenies of Lignobrycon, Rhinelepis, Spintherobolus, and Triportheus, the species-rich fish clades available at that time [22]. tribes Aspidoradini and Glandulocaudini, and the subfamilies Chiachio et al. [36] explained the division of Hypoptopoma- Cheirodontinae and Sarcoglanidinae as examples of pattern B. tinae, between lineages in the Amazon Basin (Region D) and the According to the Lagrange ancestral-area estimations, the area Brazilian East Coastal (Region A) and the Upper Parana´ (Region of the modern Atlantic Coastal Drainages (Region A) is optimized B), as the result of limited dispersal of fishes to less favourable areas as the ancestral area for three of the deepest nodes of the HNO of the continental margin. Although species of Hypoptopomatinae phylogeny. These nodes include the HNO-clade as a whole do inhabit lowland rivers in the Amazon, Orinoco and Guianas (40.8–79.7 Mya 95% HPD, mean 58.4 Mya, Fig. 5), the Hypop- regions, most species of Neoplecostominae plus Otothyrinae topomatinae (33.6–67.4 Mya 95% HPD, mean 49.9 Mya, Fig. 6), inhabit rivers and streams in the mountainous Brazilian Shield, and the Neoplecostominae and Otothyrinae (31.0–62.2 Mya 95% where they are adapted to colder and more highly oxygenated HPD, mean 45.9 Mya, Fig. 6–7). The results of the Lagrange waters [27]. Additionally, historical paleogeographic connections analysis are consistent with a river capture event at about 26.7– among the Orinoco, Amazon, and Paraguay basins are hypoth- 58.6 Mya 95% HPD (mean 42.5 Mya), allowing range expan- esized to have enabled the colonization of Hypoptopomatinae sion(s) from the Atlantic Coastal Drainages (Region A) to a region species through these basins [4]. comprised of the modern Paraguay/Lower Parana´/Uruguay Within Neoplecostominae our time-calibrated phylogeny and (Region C) and Amazon/Orinoco Basins (Region D). An Lagrange biogeographic analysis suggest a geodispersal event in

Figure 6. Partial time-calibrated tree from BEAST analysis, showing divergence ages for taxa in Hypoptopomatinae and Neoplecostominae. The curve below the phylogeny represented sea levels with the marine transgressions (MT), modified from Zachos et al. [124] and Miller et al. [125]. doi:10.1371/journal.pone.0105564.g006

PLOS ONE | www.plosone.org 10 August 2014 | Volume 9 | Issue 8 | e105564 Molecular Phylogeny and Biogeographic History of Cascudinhos

Figure 7. Partial time-calibrated tree from BEAST analysis, showing divergence ages for taxa in Otothyrinae. doi:10.1371/journal.pone.0105564.g007 the ancestral species of the clade composed of Neoplecostomus estimated in the time frame 20.7–47.4 Mya 95% HPD (mean (except N. ribeirensis) and Pareiorhina carrancas to move from the 33.29 Mya). Coastal Drainages (Region A) to the Upper Parana´ Basin (Region The results of our Lagrange analysis suggest the influence of B) at about 14.2–33.4 Mya 95% HPD (mean 22.9 Mya). Roxo river capture in the movement of Otothyrinae from Atlantic et al. [48] reported an event with a similar date in the range 15.4– Coastal Drainages (Region A) to the Amazon and Orinoco Basins 38.1 Mya 95% HPD (mean 26.7 Mya), and suggested that this (Region D). The most highly supported model (M3) of geographic geodispersal event could be a result of a headwater capture. dispersal among areas posits a connection between regions C During this time period several headwater capture events have (Paraguay, Lower Parana´ and Uruguay Basins) and D (Amazon been proposed between the Rio Tieteˆ, Rio Paraı´ba do Sul, Rio and Orinoco Basins) before 15 Mya. For more than a century Sa˜o Francisco, and Rio Ribeira de Iguape basins [16,90,91]. authors have suggested historical dispersal routs of fishes between Headwater capture is likely to have influenced ancestral fish Paraguay and Amazon basins [20,25,92–95]. These authors distributions throughout adjacent drainages, allowing the ances- suggested that most of the fish lineages represented in the tors of this group to reach the Upper Parana´ basin. Paraguay Basin can be explained by dispersal, presumably by The subfamily Otothyrinae also has a complex biogeographic means of headwater capture (geodispersal) of Amazon tributaries history among South American basins (Fig. 7). The ancestral-area (Madeira, Tocantins, Xingu) on the Brazilian Shield. However, estimation with highest ML scores gives us the origin in the geodispersal events in the reverse direction, from south to north, Coastal Drainages (Region A). The first lineage to diverge within must also be considered for taxa with origins in the La Plata and Otothyrinae (i.e. species of Corumbataia and six species of Atlantic coastal drainages, and with derived lineages in the Hisonotus, H. insperatus, H. oliveirai, H. paresi, H. piracanjuba, Amazon and Orinoco basins. Hisonotus sp. 4 and Hisonotus sp. 5), geodispersal from Coastal The Lagrange analysis also infers a river capture event affecting Drainage (Region A) to Upper Parana´ basin (Region B) and is the ancestor of the clade including New taxon sp. 1, New taxon sp.

PLOS ONE | www.plosone.org 11 August 2014 | Volume 9 | Issue 8 | e105564 Molecular Phylogeny and Biogeographic History of Cascudinhos

Figure 8. Geographic distribution and phylogeny of major clades of Hypoptopomatinae and Neoplecostominae in tropical South America. Terminal taxa may represent species or monophyletic higher taxa (see figs. 1–2 for all taxa). Numbers in parentheses after each taxon name are number of taxa sampled for each lineage. Regions: A, Coastal Drainages (Green); B, Upper Parana´ Basin (Red); C, Paraguay, Lower Parana´ and Uruguay Basins (Purple); D, Amazon and Orinoco Basins (Blue); E, Sa˜o Francisco Basin and Northeastern Drainages (Yellow). doi:10.1371/journal.pone.0105564.g008

2, Hisonotus bocaiuva, Parotocinclus cf. bahiensis, P. robustus and The results of our Lagrange analysis also indicate a geodispersal P. prata from the Atlantic Coastal Drainages (Region A) to the event from the Amazon and Orinoco basins (Region D) to the Sa˜o Sa˜o Francisco Basin and Northeastern Drainages (Region E) in the Francisco Basin and Northeastern Drainages (Region E) in the time frame 11.3–26.1 Mya 95% HPD (mean 18.2 Mya). These ancestor of Hisonotus sp. 1, Hisonotus sp. 2 and Parotocinclus aff. two regions also share extensive watershed divides with the many spilurus at about 19.3–43.2 Mya 95% HPD (mean 30.7 Mya). separate Atlantic coastal drainages of the eastern margin of the Rosa et al. [99] previously suggested that some fish species in Brazilian Shield. Ribeiro [16] suggested that the origin of the Northeastern Brazil are widespread in two or more basins, Taubate´ Graben probably resulted in the capture of several other encompassing the Sa˜o Francisco, Parnaı´ba and several adjacent adjacent rivers, such as headwaters of the Tieteˆ, Grande, Sa˜o coastal rivers basins. This is the case, for example, in Triportheus Francisco and Doce rivers. A river capture event at this signatus, Prochilodus brevis, Cichlasoma orientale and Parauche- approximate time and place is also consistent with the General nipterus galeatus. Area Cladogram of fish taxa from tropical South America [22]. 4.2. Sea-level changes as a biogeographic The results of our Lagrange analysis point to the influence of process. Periods of alternating sea-level stands can also several river-capture events permitting movements of Otothyrinae influence the distributions of lowland freshwater taxa [100–102]. lineages from the Atlantic Coastal Drainages (Region A) to the Eustatic sea-level changes under global climate controls, and Paraguay, Lower Parana´ and Uruguay Basins (Region C; Fig. 7). regional subsidence or uplift under tectonic controls, have resulted These events occurred within the group of Hisonotus species in multiple marine transgressions and regressions over the course (including the type species Hisonotus notatus), and the species of of the Cenozoic, alternately flooding and exposing low lying areas the genera Eurycheilichthys and Epactionotus. While most of the of the continental platforms, and converting lowland and coastal early-branching clades in this group inhabit the eastern margin of plains from freshwater to shallow marine ecosystems. Lo´pez- the Brazilian Shield, a few early-branching lineages occur in the Fernandez and Albert [103] identify six marine transgressions Uruguay Basin (Region C). Ribeiro [16] reported that several during the Eocene, one in the Eocene and Oligocene, three in species are shared between the isolated coastal drainages and the Oligocene and one in the Miocene, the time interval during which adjacent upland as: Cnesterodon decemmaculatus and Cnesterodon most lineages of Hypoptopomatinae diversified (Fig. 6). Ancestral brevirostratus [96], Bryconamericus patriciae [97], Hypostomus lineages of Hypoptopomatinae were present in the lowland commersoni and H. aspilogaster [98]. portions of the Amazon and Orinoco basins (Region D) from about 26.7–58.6 Mya 95% HPD (mean 42.5 Mya) to the present,

PLOS ONE | www.plosone.org 12 August 2014 | Volume 9 | Issue 8 | e105564 Molecular Phylogeny and Biogeographic History of Cascudinhos

Figure 9. Geographic distribution and phylogeny of major clades of Otothyrinae in tropical South America. Terminal taxa may represent species or monophyletic higher taxa (see figs. 1–2 for all taxa). Numbers in parentheses after each taxon name are number of taxa sampled for each lineage. Regions: A, Coastal Drainages (Green); B, Upper Parana´ Basin (Red); C, Paraguay, Lower Parana´ and Uruguay Basins (Purple); D, Amazon and Orinoco Basins (Blue); E, Sa˜o Francisco Basin and Northeastern Drainages (Yellow). doi:10.1371/journal.pone.0105564.g009 and our results suggests three events of geodispersal to the lowland Peripheral location of low-diversity, early-branching portions of the La Plata basin from about 12.6–33.1 Mya 95% lineages HPD (mean 21.7 Mya), 9.9–26.4 Mya 95% HPD (mean The ancestral-area estimations generated by Lagrange (Fig. 5) 17.0 Mya) and 0.0–10.6 Mya 95% HPD (mean 5.3 Mya) to the permit one to infer the geographic origin of the HNO-clade, and present. These populations were therefore presumably influenced of all three HNO subfamilies, in the Atlantic Coastal Drainages by numerous regional marine transgressions and regressions. (Region A), a relatively narrow strip of rivers basins that extends Marine transgressions can isolate and fragment lowland fish along the eastern continental margin. This is a reasonable populations, promoting both speciation and extinction by reducing interpretation given the disproportionately high number of low- the total amount and connectivity of freshwater habitat patches diversity, early-branching clades in all three subfamilies restricted [11,16,100,103,104]. Marine transgressions can also result in local to Region A. This interpretation also conforms to widespread population extirpations and/or allopatric speciation in upland expectations about the relative rates of macroevolutionary refugia (e.g. Albert et al. [105]. Marine regressions can expand parameters that affect net rates of diversification [111–113]. The lowland and coastal freshwater habitats, thereby promoting Lagrange DEC model of species range evolution assumes a model dispersal and reducing extinction [103,106]. of biogeographic history dominated by vicariance, in which The relatively small areal extent of river basins in the Atlantic dispersal and extinction are treated as relatively rare events coastal drainages, combined with areal expansions and contrac- [84,114]. The Lagrange model is also entirely neutral (sensu tions due to Pleistocene shoreline fluctuations, may have acted in Hubbell [115]) with respect to DEC parameter values among concert to elevate speciation and extinction rates in this region clades and regions. [107,108]. Indeed many extant fish species in the Atlantic coastal An alternative interpretation of HNO biogeographic history drainages are of high conservation concern [109]. However, the may also be considered, in which the ancestral species range was effect of Pleistocene shoreline fluctuations on fish diversity was distributed over a wide portion of southern South America in the presumably restricted to the coastal plain (areas below 100m early Cenozoic, including much of the modern Atlantic Coastal elevation), whereas most of the fish species of the Atlantic coastal (Region A), Upper Parana (Region B), and Paraguay/Lower drainages inhabit canyons in the piedmont, especially larger rivers Parana/Uruguay (Region C) areas. Under this alternative such as the Rio Doce and Ribeiro de Iguape [110]. interpretation, the accumulation of many low-diversity, early- branching clades in the Atlantic coastal drainages is expected from

PLOS ONE | www.plosone.org 13 August 2014 | Volume 9 | Issue 8 | e105564 Molecular Phylogeny and Biogeographic History of Cascudinhos patterns of diversification on landscapes with low rates of river tominae and Otothyrinae, used parametric biogeographic meth- capture. ods to estimate ancestral geographic ranges, and documented In places like the eastern margin of South America, where several historical river-capture events. The results of this study geographic range evolution is thought to have been dominated by largely support previous hypotheses on the origins and evolution of river capture [16,116–118], vicariance and geodispersal events are the major HNO-clades in space and time. The historical expected to be coupled (see the Introduction; see also Albert and biogeographic analysis of these taxa also illustrates the role of Crampton [7]). Under these conditions, rates of speciation and river capture as an important evolutionary process contributing to dispersal should be approximately matched as sources for the the extraordinary diversification of Neotropical fishes in South- introduction of new species (sensu Vellend [119]). Further, because eastern Brazil. dispersal expands species ranges, it tends to reduce extinction rates, and freshwater fish species with larger ranges generally have Supporting Information lower extinction risk [120,121]. Therefore, the combination of low speciation and extinction rates in the Atlantic coastal drainages Table S1 Species included in the present study. may have contributed to an accumulation of low-diversity clades. (DOCX) By contrast, the relatively higher rates of speciation and extinction Table S2 Summary of taxonomic information for species of in the La Plata basin is predicted to have resulted in a phylogeny Hypoptopomatinae, Neoplecostominae and Otothyrinae included with few or no low-diversity early-branching clades (see e.g. Albert in the analysis. et al. [4] fig. 2.15). In other words, regions with high species (DOC) turnover are less likely to retain low-diversity early-branching clades (i.e. the Effect Hypothesis of Vrba [122]). Table S3 Primers used in the present study to amplify partial This alternative interpretation predicts the presence of many sequences of F-reticulon 4, 16S rRNA, cytochrome oxidase low-diversity, early-branching fish lineages on landscapes with low subunit I (COI) and cytochrome B (CytB). rates of river capture. This alternative interpretation differs from (DOC) the Lagrange-generated ancestral-area estimations by positing Table S4 Nucleotide substitution models for each partition different rates of speciation and extinction in clades inhabiting the evaluated in the software PartitionFinder [68] and used in the Atlantic coastal drainages and La Plata basin. In other words, this phylogenetic analyses. *These partitions were analyzed in one alternative interpretation is not neutral with respect to DEC partition. parameter values among regions, positing instead that rates of (DOC) speciation and extinction are correlated with rates of river capture. Table S5 DEC models tested to estimate distribution ranges inherited by the descending lineages at each node of the tree. The Museums and cradles differences between the models are in the rate of dispersal among In evaluating distributional patterns of Neotropical fish distri- adjacent and no adjacent areas. * Represent the model used in the butions in southeastern Brazil, Ribeiro [16] concluded that the analysis. Atlantic coastal drainages (Region A of the present study) served as (DOC) both a cradle and a museum of diversity for different fish groups. The terms ‘‘evolutionary cradle’’ and ‘‘evolutionary museum’’ are Table S6 Substitution Saturation estimated for each gene using alternative hypotheses for the occurrence of areas with high species the index of substitution saturation (Iss) [65,66] and the rate of richness [123]. An ‘‘evolutionary cradle’’ is an area with high rates transitions/transversions evaluated in software DAMBE 5.2.31 of speciation, where environmental conditions promote speciation. [67]. By contrast, an ‘‘evolutionary museum’’ is an area with low rates (DOC) of extinction, where low rates of environmental disturbance act to Table S7 Likelihood-based tests for alternative topologies. SH preserve early-branching taxa, and where species richness and AU are probability values obtained for the Shimodaira- accumulates through long periods of geological time. Hasegawa and the Approximately Unbiased tests [76]. Asterisks The results of this study on the Hypoptopomatinae, Neople- denote significant values (P,0.05 for SH and P,0.01 for AU and costominae and Otothyrinae broadly concur with the conclusions ELW), that imply the topology is rejected. of Ribeiro [16] (Figs. 8–9). All three HNO subfamilies are inferred (DOC) by Lagrange ancestral-area estimations to have originated in the Atlantic coastal drainages, suggesting that this region served as the Acknowledgments cradle for early diversification in these clades. In addition, several lineages of Neoplecostominae remain confined to the region of the We thank Guilherme J. Costa Silva, Jefferson M. Henriques, Alex T. Atlantic coastal drainages, which therefore also appears to serve as Ferreira, Mahmoud Mehanna and Renato Devide´ for their help during the a museum for these clades. These major patterns of diversification collection expeditions; Victor A. Tagliacollo for help with the data analysis; Juan I. Montoya-Burgos and Roberto E. Reis for providing tissue samples; in Neoplecostominae in the Atlantic coastal drainages and Bruno F. Melo, Danilo Pinhal, Fla´vio C. T. Lima, Francisco Langeani Brazilian Shield were previously recognized by Roxo et al. [48]. Neto, Pedro Hollanda Carvalho, Ricardo C. Benine, Roberto E. Reis and For Hypoptopomatinae, most of the diversification occurred in Tiago P. Carvalho for fruitful discussions; Fernanda O. Martins for helps lowlands of the Amazon, Orinoco and Paraguay basins, and the with the taxonomic identification. species Lampiella gibbosa and Otocinclus affinis appears to be relictual lineages confined to the Atlantic coastal drainages. Author Contributions Diversification within Otothyrinae exhibits a pattern with Conceived and designed the experiments: FFR GSCS FF CO. Performed monophyletic lineages in each of the several regions and basins the experiments: FFR GSCS. Analyzed the data: CO FFR JSA. of the South American platform (Fig. 9). Contributed reagents/materials/analysis tools: CO. Contributed to the Here, we presented a time-calibrated phylogenetic analysis of writing of the manuscript: FFR JSA. Identified all species used in the study: the loricariid catfish subfamilies Hypoptopomatinae, Neoplecos- FFR GSCS CHZ.

PLOS ONE | www.plosone.org 14 August 2014 | Volume 9 | Issue 8 | e105564 Molecular Phylogeny and Biogeographic History of Cascudinhos

References 1. Cox CB, Moore PD (2005) Biogeography: An ecological and evolutionary 27. Langeani F (1990) Revisa˜o do geˆnero Neoplecostomus, com a descric¸a˜o de approach, 7th edn. United Kingdom: Blackwell, Oxford, pp. 498. quatro espe´cies novas do sudeste brasileiro (Ostariophysi, Siluriformes, 2. Lomolino MV, Riddle BR, Whittaker RJ, Brown JH (2010) Biogeography, 4th Loricariidae). Com Mus Cieˆnc Tec PUCRS Zool 3: 3–31. edn. Massachusetts: Sinauer Associates, Sunderland. 28. Armbruster JW (1998) Modifications of the Digestive Tract for Holding Air in 3. Vari RP, Malabarba LR (1998) Neotropical ichthyology: An overview. In: Loricariid and Scoloplacid Catfishes. Copeia 3: 663–675. Malabarba LR, Reis RE, Vari R, Lucena ZMS, Lucena CAS, editors. 29. Covain R, Fisch-Muller S (2007) The genera of the Neotropical armored Phylogeny and Classification of Neotropical Fishes. Edipucrs, Porto Alegre. pp. catfish subfamily Loricariinae (Siluriformes: Loricariidae): a practical key and 1–11. synopsis. Zootaxa 1462: 1–40. 4. Albert JS, Petry P, Reis RE (2011) Major biogeographic and phylogenetic 30. Nelson JA, Wubah DA, Whitmer ME, Johnson EA, Stewart DJ (1999) Wood- patterns.In:AlbertJS,ReisRE,editors.HistoricalBiogeographyof eating catfishes of the genus Panaque: gut microflora and cellulolytic enzyme Neotropical Freshwater Fishes. University of California Press, Berkeley, Los activities. J Fish Biol 54: 1069–1082. Angeles. pp. 21–57. 31. Sabaj MH, Armbruster JW, Page LM (1999) Spawning in Ancistrus 5. Waters JM, Allibone RM, Wallis GP (2006) Geological subsidence, river (Siluriformes: Loricariidae) with comments on the evolution of snout tentacles capture, and cladogenesis of galaxiid fish lineages in central New Zealand. as a novel reproductive strategy: larval mimicry. Ichthyol Explor Freshwaters Biol J Linn Soc 88: 367–376. 10: 217–229. 6. Winemiller KO, Lo´pez-Ferna´ndez H, Taphorn DC, Nico LG, Duque AB 32. Armbruster JW (2004) Phylogenetic relationships of the sucker-mouth armored (2008) Fish assemblages of the Casiquiare River, a corridor and zoogeograph- catfishes (Loricariidae) with particular emphasis on the Hypostominae and the ical filter for dispersal between the Orinoco and Amazon basins. J Biogeo- Ancistrinae. Zool J Linn Soc 141: 1–80. graphy 35: 1551–1563. 33. Bizerril CRSF (1994) Ana´lise taxonoˆmica e biogeogra´fica da ictiofauna de a´gua 7. Albert JS, Crampton WGR (2010) The geography and ecology of doce do leste brasileiro. Acta Biol Leopoldensia 16: 51–80. diversification in Neotropical freshwaters. Nature Education Knowledge 1: 34. Schaefer SA (1991) Phylogenetic Analysis of the Loricariid Subfamily 13–19. Hypoptopomatinae (Pisces, Siluroidei, Loricariidae), with Comments on 8. Smith GR (1981) Late Cenozoic freshwater fishes of North America. Annu Rev Generic Diagnoses and Geographic-Distribution. Zool J Linn Soc 102: 1–41. Ecol Syst 12: 163–193. 35. Schaefer SA (1998) Conflict and resolution: impact of new taxa on phylogenetic 9. Hocutt CH, Wiley EO (1986) Zoogeography of the Freshwater Fishes of North studies of the Neotropical cascudinhos (Siluroidei: Loricariidae). In: Malabarba America. New York: John Wiley and Sons. LR, Reis RE, Vari RP, Lucena ZMS, Lucena CAS, editors. Phylogeny and 10. Mayden RL (1988) Vicariance biogeography, parsimony, and evolution in classification of neotropical fishes. Porto Alegre: Edipucrs. pp. 375–400. North American freshwater Fishes. Syst Zool 37: 329–355. 36. Chiachio MC, Oliveira C, Montoya-Burgos JI (2008) Molecular systematic and 11. Lundberg JG, Marshall LG, Guerrero J, Horton B, Malabarba MCSL, et al. historical biogeography of the armored Neotropical catfishes Hypoptopoma- (1998) The stage for Neotropical fish diversification: A history of tropical South tinae and Neoplecostominae (Siluriformes: Loricariidae). Mol Phyl Evol 49: American rivers. In: Malabarba LR, Reis RE, Vari RP, Lucena ZMS, Lucena 606–617. CAS, editors. Phylogeny and Classification of Neotropical Fishes. Porto Alegre: 37. Cramer CA, Bonatto SL, Reis RE (2011) Molecular phylogeny of the Edipucrs. pp. 13–48. Neoplecostominae and Hypoptopomatinae (Siluriformes: Loricariidae) using multiple genes. Mol Phyl Evol 59: 43–52. 12. Almeida FFM, Carneiro CDR (1998) Origem e evoluc¸a˜o da Serra do Mar. Rev 38. Cramer CA, Liedke AMR, Bonatto LS, Reis RE (2008) The phylogenetic Bras Geo 28: 135–150. relationship of the Hypoptopomatinae and Neoplecostominae (Siluriformes: 13. Bishop P (1995) Drainage rearrangement by river capture, beheading and Loricariidae) as inferred from mitochondrial cytochrome c oxidase I sequences. diversion. Progr Phys Geogr 19: 449–473. Bull Fish Biol 9: 51–59. 14. Wilkinson MJ, Marshall LG, Lundberg JG (2006) River behavior on megafans 39. Eschmeyer WN, Fong JD (2013) Species by Family/Subfamily. Available: and potential influences on diversification and distribution of aquatic http://research.calacademy.org/research/ichthyology/catalog/SpeciesByFamily. organisms. J South Am Earth Sci 21: 151–172. asp. Accessed 18 November 2013. 15. Wilkinson MJ, Marshall LG, Lundberg JG, Kreslavsky MH (2010) Megafan 40. Eigenmann CH, Eigenmann RS (1890) South America Nematognathi. PhD environments in northern South America and their impact on Amazon Thesis, San Francisco: California Academy of Science. Neogene aquatic ecosystems. In: Horn C, Wesselingh EP, editors, Amazonia, 41. Regan CT (1904) A monograph of the fishes of the family Loricariidae. Trans Landscape and Species Evolution: A Look into the Past, Blackwell Publishing. Zool Soc Lond 17: 191–350. pp. 165–185. 42. Gosline WA (1947) Contributions to the classification of the loricariid catfishes. 16. Ribeiro AC (2006) Tectonic history and the biogeography of the freshwater Arq Mus Nac, Rio de Janeiro 41: 79–134. fishes from the coastal drainages of eastern Brazil: an example of faunal 43. Isbru¨cker IJH (1980) Classification and catalogue of the mailed Loricariidae evolution associated with a divergent continental margin. Neotr Ichthyology 4: (Pisces, Siluriformes). Inst v Tax Zoo¨l (Zoo¨l Mus) Univ Amsterdam 22: 1–181. 225–246. 44. Howes GJ (1983) The cranial muscles of loricarioid catfishes, their homologies 17. Grant CEH, Lowe WH, Fagan WF (2007) Living in the branches: population and value as taxonomic characters (Teleostei: Siluroidei). Bull br Mus nat Hist dynamics and ecological processes in dendritic networks. Ecology Letters 10: Zool 45: 309–345. 165–175. 45. Schaefer SA (1987) Osteology of Hypostomus plecostomus (Linnaeus) with a 18. Muneepeerakul R, Bertuzzo E, Lynch HJ, Fagan WF, Rinaldo A, et al. (2008) phylogenetic analysis of the loricariids subfamilies (Pisces: Siluroidei). Nat Hist Neutral metacommunity models predict fish diversity patterns in Mississippi– Mus Los Angeles 394: 1–31. Missouri basin. Nature 453: 220–222. 46. Montoya-Burgos JI, Muller S, Weber C, Pawlowski J (1998) Phylogenetic 19. Bertuzzo E, Muneepeerakul R, Lynch HJ, Fagan WF, Rodriguez-Iturbe I, relationships of the Loricariidae (Siluriformes) based on mitochondrial rRNA et al. (2009) On the geographic range of freshwater fish in river basins. Water gene sequences. In: Malabarba LR, Reis RE, Vari RP, Lucena ZM, Lucena Resour Res 45(11). CAS, editors. Phylogeny and classification of Neotropical fishes. Porto Alegre: 20. Carvalho TP, Albert JS (2011) The Amazon-Paraguay divide. In: Albert JS, Edipucrs. pp. 363–375. Reis RE, editors. Historical Biogeography of Neotropical Freshwater Fishes. 47. Reis RE, Pereira EHL, Armbruster JW (2006) , a new loricariid University of California Press, Berkeley, California. pp. 193–202. catfish subfamily (Teleostei, Siluriformes), with revisions of Delturus and 21. Lieberman BS (2008) Emerging syntheses between palaeobiogeography and Hemipsilichthys. Zool J Linn Soc 147: 277–299. macroevolutionary theory. Proc R Soc Vic 120: 51–57. 48. Roxo FF, Zawadzki CH, Alexandrou MA, Costa Silva GJ, Chiachio MC, et al. 22. Albert JS, Carvalho TP (2011) Neogene assembly of modern faunas. In: Albert (2012) Evolutionary and biogeographic history of the subfamily Neoplecosto- JS, Reis RE, editors. Historical Biogeography of Neotropical freshwater fishes. minae (Siluriformes: Loricariidae). Ecology and Evolution 2: 2438–2449. Los Angeles: University of California Press, Berkeley. pp. 119–136. 49. Roxo FF, Zawadzki CH, Costa Silva GJ, Chiachio MC, Foresti F, et al. (2012) 23. Reis RE, Kullander SO, Ferraris CJ (2003) CLOFFSCA-Check list of the Molecular systematics of the armored neotropical catfish subfamily Neople- freshwater fishes of South and Central America. Edipucrs. costominae (Siluriformes, Loricariidae). Zootaxa 3390: 33–42. 24. Abell R, Thieme ML, Revenga C, Bryer M, Kottelat M, et al. (2008) 50. Arratia G (1987) Description of the primitive family Diplomystidae (Silur- Freshwater ecoregions of the world: A new map of biogeographic units for iformes, Teleostei, Pisces): morphology, and phylogenetic implica- freshwater biodiversity conservation. Bioscience 58: 403–414. tions. Bonn Zool Monogr 24: 1–120. 25. Ribeiro AC, Jacob RM, Silva RRSR, Lima FCT, Ferreira DC, et al. (2013) 51. de Pinna MCC (1993) Higher-level phylogeny of Siluriformes (Teleostei, Distributions and phylogeographic data of rheophilic freshwater fishes provide Ostariophysi), with a new classification of the order. PhD Thesis, New York: evidences on the geographic extension of a central-Brazilian Amazonian City University of New York. palaeoplateau in the area of the present day Pantanal Wetland. Neotr 52. de Pinna MCC (1998) Phylogenetic relationships of Neotropical Siluriformes Ichthyology 11: 319–326. (Teleostei: Ostariophysi): historical overview and synthesis of hypotheses. In: 26. Eschmeyer W (2013) Catalog of fishes. Electronic publication in ‘‘World Wide Malabarba LR, Reis RE, Vari RP, Lucena CAS, Lucena ZMS, editors. Web’’. Available: http://www.calacademy.org/research/ichthyology/catalog. Phylogeny and Classification of Neotropical Fishes. Mus Cieˆnc Tecnol Accessed 20 February 2013. PUCRS, Porto Alegre. pp. 279–330.

PLOS ONE | www.plosone.org 15 August 2014 | Volume 9 | Issue 8 | e105564 Molecular Phylogeny and Biogeographic History of Cascudinhos

53. Grande L (1987) Redescription of Hypsidoris farsonensis (Teleostei: Silur- 84. Ree RH, Smith SA (2008) Maximum likelihood inference of geographic range iformes), with a reassessment of its phylogenetic relationships. J Vert Paleont 7: evolution by dispersal, local extinction, and cladogenesis. Syst Biol 57: 4–14. 24–54. 85. Reis RE, Schaefer SA (1998) New cascudinhos from southern Brazil: 54. Grande L, de Pinna MCC (1998) Description of a second species of Hypsidoris systematics, endemism, and relationships (Siluriformes, Loricariidae, Hypopto- and a reevaluation of the genus and family Hypsidoridae. J Vert Paleont 18: pomatinae). Am Mus Novit 3254: 1–25. 451–474. 86. Pitman WC III, Cande SC, LaBrecque J, Pindell J (1993) Fragmentation of 55. Mo T (1991) Anatomy, relationships and systematics of the Bagridae (Teleostei, Gondwana: The separation of Africa from South America. In: Goldblatt P, Siluroidei) with a hypothesis of siluroid phylogeny. Theses Zoologicae 17, editor. Biological Relationships between Africa and South America. Connecti- Koenigstein: Koeltz Scientific Books. cut: Yale University Press, New Haven. pp. 15–34. 56. Sullivan JP, Lundberg JG, Hardman M (2006) A phylogenetic analysis of the 87. Cesero P, Ponte FC (1997) Ana´lise comparativa da paleogeologia dos litorais major groups of catfishes (Teleostei: Siluriformes) using rag1 and rag2 nuclear atlaˆnticos brasileiro e africano. Bol Geocieˆnc PETROBRAS 11. pp. 1–18. gene sequences. Mol Phyl Evol 41: 636–662. 88. Scotese CR (2004) Cenozoic and Mesozoic paleogeography: Changing 57. Kocher TD, Thomas WK, Meyer A, Edwards SV, Pa¨bo S, et al. (1989) terrestrial pathways. In: Lomolino MV, Heaney LR, editors. Frontiers of Dynamics of mitochondrial DNA evolution in animals: Amplification and Biogeography. Massachusetts: Sinauer Associates, Sunderland. pp. 9–26. sequencing with conserved primers. PNAS 86: 6196–6200. 89. Blakey R (2006) Plate tectonics and continental drift: Regional paleogeographic 58. Oliveira C, Avelino GS, Abe KT, Mariguela TC, Benine RC, et al. (2011) views of earth history. Available: http://jan.ucc.nau.edu/,rcb7/globaltext. Phylogenetic relationships within the speciose family Characidae (Teleostei: html. Accessed 14 August 2013. Ostariophysi: Characiformes) based on multilocus analysis and extensive 90. Ab’Saber AN (1957) O problema das conexo˜es antigas e da separac¸a˜o da ingroup sampling. BMC Evol Biol 11: 275. drenagem do Paraı´ba e Tieteˆ. Bol Paulista Geografia 26: 38–49. 59. Ward RD, Zemlak TS, Innes BH, Last PR, Hebert PDN (2005) DNA 91. Ab’Saber AN (1998) Megageomorfologia do Territo´rio Brasileiro. In: Cunha barcoding Australia’s fish species. Phil Trans R Soc B, 360: 1847–1857. SB, Guerra AJT, editors. Geomorfologia do Brasil, Rio de Janeiro: Bertrand. 60. Hall TA (1999) BioEdit: a user-friendly biological sequence alignment editor pp. 71–106. and analysis program for Windows 95/98/NT. Nucleic Acids Symp Ser 41: 92. Eigenmann CH, Eigenmann RS (1891) A catalogue of the freshwater fishes of 95–98. South America. Proc U S Nat Mus 14: 1–81. 61. Edgar RC (2004) Muscle: a multiple sequence alignment method with reduced 93. Jordan DS (1896) Science sketches. A.C. McClurg and Company, Chicago. time and space complexity. BMC Bioinformatics 5: 1–19. 94. Eigenmann CH (1906) The freshwater fishes of the South and Middle America. 62. Stamatakis A, Hoover P, Rougemont J (2008) A rapid bootstrap algorithm for PopSci 68: 515–530. the RAxML web servers. Syst Biol 57: 758–771. 95. Pearson NE (1937) The fishes of the Beni-Mamore´ and Paraguay basin, and a 63. Stamatakis A (2006) RAxML-VI-HPC: Maximum likelihood-based phyloge- discussion of the origin of the Paraguayan fauna. Proc Calif Acad Sci 23: 99– netic analyses with thousands of taxa and mixed models. Bioinformatics 22: 114. 2688–2690. 96. Lucinda PHF (2005) Systematics of the genus Cnesterodon Garman, 1895 64. Tamura K, Peterson D, Peterson N, Stecher G, Nei M, et al. (2011) MEGA5: (Cyprinodontiformes: Poeciliidae: Poeciliinae). Neotr Ichthyology 3: 259–270. Molecular Evolutionary Genetics Analysis using Maximum Likelihood, 97. Silva JFP (2004) Two new species of Bryconamericus Eigenmann (Characi- Evolutionary Distance, and Maximum Parsimony Methods. Mol Biol Evol formes: Characidae) from southern Brazil. Neotr Ichthyology 2: 55–60. 28: 2731–2739. 98. Reis RE, Weber C, Malabarba LR (1990) Review of the genus Hypostomus 65. Xia X, Xie Z, Salemi M, Chen L, Wang Y (2003) An index of substitution Lace´pe`de, 1803 from southern Brazil, with descriptions of three new species saturation and its application. Mol Phyl Evol 26: 1–7. (Pisces, Siluriformes, Loricariidae). Rev Sui Zoo 97: 729–766. 66. Xia X, Lemey P (2009) Assessing substitution saturation with DAMBE. In: 99. Rosa RS, Menezes NA, Britski HA, Costa WJEM, Groth F (2004) Diversidade, Lemey P, Salemi M, Vandamme AM, editors. The Phylogenetic Handbook: A padro˜es de distribuic¸a˜o e conservac¸a˜o dos peixes da Caatinga. In: Leal IR, Practical Approach to DNA and Protein Phylogeny. Cambridge University Tabarelli M, Silva JMC, editors. Ecologia e conservac¸a˜o da Caatinga. Recife: Press. pp. 615–630. Edufpe. pp. 135–180. 67. Xia X, Xie Z (2001) DAMBE: Data analysis in molecular biology and 100. Lovejoy NR, Albert JS, Crampton WGR (2006) Miocene marine incursions evolution. J Hered 92: 371–373. and marine/freshwater transitions: Evidence from Neotropical fishes. J South 68. Lanfear R, Calcott B, Ho SYW, Guindon S (2012) PartitionFinder: combined Am Earth Sci 21: 5–13. selection of partitioning schemes and substitution models for phylogenetic 101. Albert JS, Reis RE (2011) Historical Biogeography of Neotropical freshwater analyses. Mol Biol Evol 29: 1695–1701. fishes, 1st edn. Los Angeles: University of California Press, Berkeley. 69. Felsenstein J (1985) Confidence limits on phylogenies: an approach using the 102. Bloom DD, Lovejoy NR (2011) The biogeography of marine incursions in bootstrap. Evolution 39: 783–791. South America. In: Albert JS, Reis RE, editors. Historical Biogeography of 70. Huelsenbeck JP, Ronquist F (2001) MrBayes: Bayesian inference of Neotropical freshwater fishes. Los Angeles: University of California Press, phylogenetic trees. Bioinformatics 17: 754–755. Berkeley. pp. 137–144. 71. Ronquist F, Huelsenbeck JP (2003) MrBayes 3: Bayesian phylogenetic 103. Lo´pez-Ferna´ndez H, Albert JS (2011) Paleogene Radiations. In: Albert JS, Reis inference under mixed models. Bioinformatics 19: 1572–1574. RE, editors. Historical Biogeography of Neotropical freshwater fishes. Los 72. Rambaut A, Drummond AJ (2007a) Tracerv1.5. Available: http://beast.bio. Angeles: University of California Press, Berkeley. pp. 105–118. ed.ac.uk/Tracer. Accessed 04 November 2013. 104. Sabaj-Pe´rez MH, Orangel AS, Lundberg JG (2007) Fossil catfishes of the 73. Swofford DL (2003) PAUP*: phylogenetic analysis using parsimony (*and other families Doradidae and Pimelodidae (Teleostei: Siluriformes) from the Miocene methods). Version 4. Massachusetts: Sinauer Associates, Sunderland, MA. Urumaco Formation of Venezuela. Proc Acad Natural Sci Phil 156: 157–194. 74. Jobb G, von Haeseler A, Strimmer K (2004) TREEFINDER: a powerful 105. Albert JS, Lovejoy NR, Crampton WGR (2006) Miocene tectonism and the graphical analysis environment for molecular phylogenetics. BMC Evol Biol 4: separation of cis-and trans-Andean river basins: Evidence from Neotropical 4–18. fishes. J South Am Earth Sci 21: 14–27. 75. Shimodaira H, Hasegawa M (1999) Multiple comparisons of log-likelihoods 106. Lo´pez-Ferna´ndez H, Arbour JH, Winemiller KO, Honeycutt RL (2013) with applications to phylogenetic inference. Mol Biol Evol 16: 1114–1116. Testing for ancient adaptive radiations in Neotropical cichlid fishes. Evolution 76. Shimodaira H (2002) An approximately unbiased test of phylogenetic tree 67: 1321–1337. selection. Syst Biol 51: 492–508. 107. Beheregaray LB, Sunnucks P, Briscoe DA (2002) A rapid fish radiation 77. Strimmer K, Rambaut A (2002) Inferring confidence sets of possibly associated with the last sea-level changes in southern Brazil: the silverside misspecified gene trees. Proc R Soc B 269: 137–142. Odontesthes perugiae complex. Proc R Soc Lond B: Biological Sciences 269: 78. Lundberg JG (1993) African South America freshwater fish clade and 65–73. continental drift: problems with a paradigm. In: Goldblatt P, editor. The 108. do Amaral PG (2012) Paleoenvironmental reconstruction of a Late Quaternary Biotic Relationship between Africa and South America. Yale University Press. lagoon system in southern Brazil (Jagurana region, Santa Catarina state) based pp. 156–199. on multi-proxy analysis. Quaternary International 279: 120. 79. Lundberg JG, Sullivan JP, Rodiles-Hernandez R, Hendrickson DA (2007) 109. Reis RE (2013) Conserving the freshwater fishes of South America. Discovery of African roots for the Mesoamerican Chiapas catfish, Lacantunia International Zoo Yearbook 47: 65–70. enigmatica, requires an ancient intercontinental passage. Proc Acad Nat Sci 110. Malabarba LR, Neto PC, Bertaco VA, Carvalho TP, dos Santos JF, et al. Phil 156: 39–53. (2013) Guia de identificac¸a˜o dos peixes da bacia do rio Tramandaı´. Porto 80. Cockerell TDA (1925) A fossil fish of the family Callichthyidae. Science 62: Alegre, Via Sapiens. 397–398. 111. Bloom DD, Weir JT, Piller KR, Lovejoy NR (2013) Do freshwater fishes 81. Marshall LG, Sempere T, Butler RF (1997) Chronostratigraphy of the diversify faster than marine fishes? A test using state-dependent diversification mammal-bearing Paleocene of South America. J South Am Earth Sci 10: 49– analyses and molecular phylogenetics of new world silversides (Atherinopsidae). 70. Evolution 67: 2040–2057. 82. Rambaut A, Drummond AJ (2007b) TreeAnnotator v1.7.5. Available: http:// 112. Pyron RA, Burbrink FT (2013) Phylogenetic estimates of speciation and beast.bio.ed.ac.uk/TreeAnnotator. Accessed 09 November 2013. extinction rates for testing ecological and evolutionary hypotheses. Trends Ecol 83. Ree RH, Moore BR, Webb CO, Donoghue MJ (2005) A likelihood framework Evol 28: 729–736. for inferring the evolution of geographic range on phylogenetic trees. Evolution 113. Rabosky DL (2013) Diversity Dependence, Ecological Speciation, and the Role 59: 2299–2311. of Competition in Macroevolution. Annu Rev Ecol Evol Syst 44(1).

PLOS ONE | www.plosone.org 16 August 2014 | Volume 9 | Issue 8 | e105564 Molecular Phylogeny and Biogeographic History of Cascudinhos

114. Ronquist F, Sanmartı´n I (2011) Phylogenetic methods in biogeography. Annu 119. Vellend M (2010) Conceptual synthesis in community ecology. Q Rev Biol 85: Rev Ecol Evol Syst 42: 441–464. 183–206. 115. Hubbell SP (2001) The Unified Neutral Theory of Biodiversity and 120. Fagan WF (2002) Connectivity, fragmentation, and extinction risk in dendritic Biogeography Monographs in Population Biology v32. New Jersey: Princeton metapopulations. Ecology 83: 3243–3249. University Press. 121. O’Grady JJ, Reed DH, Brook BW, Frankham R (2004) What are the best 116. Buckup PA (2011) The Eastern Brazilian Shield. In: Albert JS, Reis RE, correlates of predicted extinction risk? Biol Conserv 118: 513–520. editors. Historical Biogeography of Neotropical Freshwater Fishes. California: 122. Vrba ES (1984) What is species selection? Syst Zool 33: 318–328. University of California Press, Berkeley. pp. 203–210. 123. Stebbins GL (1974) Flowering Plants: Evolution above the Species Level. 117. Lima FC, Ribeiro AC (2011) Continental-scale tectonic controls of biogeog- Massachusetts: Belknap Press of Harvard University Press, Cambridge. raphy and ecology. In: Albert JS, Reis RE, editors. Historical biogeography of 124. Zachos J, Pagani M, Sloan L, Thomas E, Billups K (2001) Trends, rhythms, Neotropical freshwater fishes. Los Angeles: University of California Press, and aberrations in global climate 65 Mya to present. Science 292: 686–693. Berkeley. pp. 145–164. 125. Miller KG, Kominz CAS, Browning JV, Wright DJ, Mountain GS, et al. 118. Pereira TL, Santos U, Schaefer CE, Souza GO, Paiva SR, et al. (2012) (2005) The Phanerozoic record of global sea-level change. Science 310: 1293– Dispersal and vicariance of Hoplias malabaricus (Bloch, 1794) (Teleostei, 1298. Erythrinidae) populations of the Brazilian continental margin. J Biogeogr 40: 905–914.

PLOS ONE | www.plosone.org 17 August 2014 | Volume 9 | Issue 8 | e105564